Research on Measurement and Application of China’s Regional Logistics Development Level under Low Carbon Environment
Abstract
:1. Introduction
2. Theoretical Basis
2.1. Entropy Weighting Method
- Standardized processing of data: assume that m evaluation objects, n evaluation signals, get the original evaluation, , make
- Calculation of weights for each indicator:
- Calculation of entropy for each indicator:
- Determination of weights for each indicator:
2.2. Cloud Models
2.2.1. The Cloud Models
2.2.2. Numerical Characteristics of Clouds
2.2.3. Cloud Generator
2.3. Carbon Emission Measurement
3. Regional Logistics Decarbonization Development Evaluation Model Construction
3.1. Evaluation Index System for Low-Carbon Development of Regional Logistics
3.2. Construction of Evaluation Model
3.2.1. Defining the Object and Domain of Cloud Model Evaluation
3.2.2. Settle the Evaluation Level of Each Indicator
3.2.3. Decide the Cloud Numerical Eigenvalues of Each Evaluation Index and Cloud Model Map
3.2.4. Determine the Affiliation of Each Evaluation Index
3.2.5. Entropy Weighting Method to Assign the Index Weights
3.2.6. Determine the Comprehensive Evaluation Level of Regional Logistics Decarbonization Development
4. Empirical Analysis and Pathway Study
4.1. Data Sources and Carbon Emission Measurement in Beijing, Tianjin and Hebei
4.1.1. Data Sources
4.1.2. Carbon Emission Measurement in Beijing, Tianjin and Hebei
4.2. Evaluation of the Low Carbon Development of Logistics in Beijing, Tianjin and Hebei
4.2.1. Selection of Indicator Samples
4.2.2. Determine the Level of Each Evaluation Index
4.2.3. Determine the Cloud Digital Characteristic Value of Each Evaluation Index and Cloud Model Map
4.2.4. Calculate the Affiliation Degree of Each Index
4.2.5. Entropy Weighting Method to Determine the Weights
4.2.6. Determine the Comprehensive Evaluation Level of Regional Logistics Index
4.3. Determination of Influencing Factors and Suggestions for Countermeasures
4.3.1. Determination of Influencing Factors
4.3.2. Suggestions for Countermeasures to the Low-Carbon Development of Logistics in Beijing-Tianjin-Hebei Region
- (1)
- From the shortcomings of the development of low-carbon logistics in Beijing, Tianjin and Hebei in recent years, Beijing needs to strengthen two aspects of low-carbon logistics strength and low-carbon logistics potential, especially the three modules of logistics industry efficiency, logistics industry input and demand, and technical support. Tianjin should start with a balanced approach to logistics industry efficiency, input, output, demand and technical support in order to improve the overall low-carbon development of logistics. Hebei Province should strengthen the development of logistics economy, improve the practice base of logistics enterprises, promote industrial clusters and create a logistics ecological chain while improving economic strength, so as to enhance the level of logistics low carbonization in all aspects.
- (2)
- Strengthen the division of labor and cooperation between Beijing, Tianjin and Hebei in logistics. In the 13th Five-Year Plan, Beijing, Tianjin and Hebei are planned as a whole region, and the respective positions of the three provinces and cities have been clarified. In this context, the logistics industry synergy among the three provinces and cities should optimize the logistics network and divide the work according to the characteristics of each region. Beijing gives full play to the advantages of science and technology and innovates the development of logistics industry while improving the consumer-oriented end logistics system. Tianjin focuses on building a port logistics base in the context of the linkage of three ports. Compare with Beijing and Tianjin, Hebei Province is rich in resources, so it should undertake the transfer of Beijing-Tianjin trade logistics and build Hebei into an important base for modern trade logistics in the country.
- (3)
- The government increases the policy support for developing low-carbon logistics. The development of low-carbon logistics in Beijing, Tianjin and Hebei needs the cooperation and joint planning of the three regions, and government departments should give support and guidance in policies, such as encouraging the development of ecological logistics industry chain, providing relevant enterprises with corresponding technical support or improving the reasonableness of taxation and financing policy preferences, etc. In addition, while developing regional logistics and economy at high-speed, we should actively promote the idea of green logistics, change the traditional concept of consumers, advocate low-carbon consumption and raise the low-carbon awareness of the logistics industry.
- (4)
- Improve the level of informatization of low-carbon logistics in Beijing, Tianjin and Hebei. Informatization is an important feature of modern logistics and an effective way to achieve low carbon regional logistics. In the process of integrated development and communication, Beijing, Tianjin and Hebei provinces and cities should break the information silos, establish and improve the logistics information exchange platform, and share and freely exchange logistics information so as to connect the information of each node of the supply chain and give full play to the advantages of regional informatization, to reduce logistics costs and improve logistics efficiency.
5. Conclusions
- (1)
- Twenty-one indicators are selected from the three dimensions of low-carbon logistics environment support, low-carbon logistics strength and low-carbon logistics potential to establish the regional logistics low-carbonization development evaluation index system. Combined with the cloud model and entropy weight method to build the index evaluation model, which solves the problem of fuzziness and randomness in the process of regional logistics low-carbonization development evaluation.
- (2)
- The evaluation model of regional logistics decarbonization development can show the development changes of each region in spatial and temporal dimensions and also solve the problem of horizontal comparison between different regions, giving quantitative results of different regions and different times. Then, according to the quantitative results, can discover the shortcomings of regional logistics decarbonization development and provide theoretical support for the further development of regional logistics.
- (3)
- The entropy weight-cloud model method uses the characteristic indicators that can reflect the complex relationship between multiple factors to derive the corresponding evaluation level. It makes the evaluation results more intuitive and accurate through the cloud diagram and calculation of evaluation level. At the same time, it provides reference for the shortcomings of regional logistics decarbonization development, which is of positive significance to enhance the development of regional logistics decarbonization.
- (4)
- The development of regional logistics decarbonization is a complex and continuously changing process, and future research can further improve the evaluation index system, optimize the evaluation model, and enhance the accuracy and applicability of the evaluation model.
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Butner, K.; Geuder, D.; Hittner, J. Mastering Carbon Management: Balancing Trade-Offs to Optimize Supply Chain Efficiencies; IBM Institute for Business Value: New York, NY, USA, 2008; pp. 2–12. [Google Scholar]
- Dada, A.; Staake, T. Carbon footprints from enterprises to product instances: The potential of the EPC network. In Beherrschbare Systeme-Dank Informatik; Jahrestagung der Gesellschaft für Informatik: München, Germany, 2008; pp. 873–878. [Google Scholar]
- Piecyk, M.I.; Mc Kinnon, A.C. Forecasting the carbon footprint of road freight transport in 2020. Int. J. Prod. Econ. 2010, 128, 31–42. [Google Scholar] [CrossRef]
- Wang, L.P.; Liu, M.H. Research on carbon emission measurement and influencing factors of China’s logistics industry based on input-output method. Resour. Sci. 2018, 40, 195–206. [Google Scholar]
- Liu, Y.; Li, L. Study on decoupling and influencing factors of carbon emissions from logistics industry in China. Environ. Sci. Technol. 2018, 41, 177–181. [Google Scholar]
- Timilsina, G.R.; Shrestha, A. Transport sector CO2 emissions growth in Asia: Underlying factors and policy options. Energy Policy 2009, 37, 4523–4539. [Google Scholar] [CrossRef]
- Yang, L.; Cai, Y.; Zhong, X.; Shi, Y.; Zhang, Z. A Carbon Emission Evaluation for an Integrated Logistics System—A Case Study of the Port of Shenzhen. Sustainability 2017, 9, 462. [Google Scholar] [CrossRef] [Green Version]
- Yang, Y.W.; Wu, A.L.; Zhu, Y.Y. Decomposition and dynamic simulation of carbon emission drivers: Taking Inner Mongolia autonomous region as an example. Stat. Decis. Mak. 2020, 36, 76–80. [Google Scholar]
- Li, F.G.; Wu, L.J. Decomposition of carbon emission drivers based on LMDI method. Stat. Decis. Mak. 2019, 35, 101–104. [Google Scholar]
- Men, D.; Huang, X. Study on influencing factors of carbon emission in Jiangxi Province—Based on LMDI decomposition method. Ecol. Econ. 2019, 35, 31–35. [Google Scholar]
- Wehner, J. Energy Effificiency in Logistics: An Interactive Approach to Capacity Utilisation. Sustainability 2018, 10, 1727. [Google Scholar] [CrossRef] [Green Version]
- Peng, L. Evaluation of low carbon logistics capability of logistics enterprises in China. Bus. Econ. Res. 2017, 10, 87–89. [Google Scholar]
- Xiang, C.; Gong, B.G.; Wu, B.L. Low carbon behavior capability evaluation model of logistics service supply chain. Stat. Decis. Mak. 2017, 5, 67–71. [Google Scholar]
- Jian, L.; Jian, T. Research on the impact of China’s industrial structure adjustment on low carbon logistics efficiency—An empirical analysis based on super efficiency DEA low carbon logistics efficiency evaluation model. Price Theory Pract. 2017, 12, 130–133. [Google Scholar]
- Yang, C.M. Efficiency measurement of Jiangsu logistics industry under low carbon constraints. East China Econ. Manag. 2018, 32, 27–32. [Google Scholar]
- Chaabane, A.; Ramudhin, A.; Paquet, M. Design of sustainable supply chains under the emission trading scheme. Int. J. Prod. Econ. 2012, 135, 37–49. [Google Scholar] [CrossRef]
- Li, C. Research on the development status and countermeasures of low carbon logistics in China. Logist. Eng. Manag. 2018, 40, 1–3. [Google Scholar]
- Dong, Y.X. Research on the development status and trend of China’s low carbon logistics from an international perspective. Price Mon. 2018, 12, 46–49. [Google Scholar]
- Zhang, H. Analysis of low carbon logistics development countermeasures from the perspective of Beijing Tianjin Hebei synergy. Reform Strategy 2016, 32, 118–120. [Google Scholar]
- Zhao, S.L.; Yang, X.Y.; Song, W. General strategy of transportation and logistics integration in Beijing, Tianjin and Hebei from the perspective of low-carbon economy. China Stat. 2017, 12, 6–8. [Google Scholar]
- Ma, Y. Research on Total Factor Productivity of China’s Logistics Industry under Low Carbon Constraints; Northeast University of Finance and Economics: Dalian, China, 2014. [Google Scholar]
- Fei, X. Research on the Coordinated Development of Logistics Industry and Economic Decarbonization; Nanchang University: Nanchang, China, 2015. [Google Scholar]
- Gao, F. Research on the Coordinated Development of China’s Logistics Industry and Low-Carbon Economy; Tianjin University of Technology: Tianjin, China, 2016. [Google Scholar]
- Qin, Y. Analysis of Regional Logistics Efficiency and Its Influencing Factors under Low Carbon Economy; East China Jiaotong University: Nanchang, China, 2015. [Google Scholar]
- Lina, S. Research on the Difference of Low-Carbon Logistics Performance of Regions along the Silk Road Economic Belt in China; Zhengzhou University: Zhenzhou, China, 2018. [Google Scholar]
- Wang, X. Research on the Mechanism and Path of Low-Carbon Development of Regional Logistics; Tianjin University: Tianjin, China, 2017. [Google Scholar]
- Li, D.Y. Uncertain Artificial Intelligence; National Defense Industry Press: Beijing, China, 2005. [Google Scholar]
- Wang, L.; Zhao, H.; Liu, X.; Zhang, Z.L.; Xu-Hui Steve, E. Optimal Remanufacturing Service Resource Allocation for Generalized Growth of Retired Mechanical Products: Maximizing Matching Efficiency. IEEE Access 2021, 9, 89655–89674. [Google Scholar]
- Jia, L.; Yu, Y.; Li, Z.; Qin, S.; Guo, J.; Zhang, Y.; Jin, Y. Study on the Hg0 removal characteristics and synergistic mechanism of iron-based modified biochar doped with multiple metals. Bioresour. Technol. 2021, 332, 125086. [Google Scholar] [CrossRef] [PubMed]
- Xu, Q.W.; Xu, K.L. Evaluation of ambient air quality based on synthetic cloud model. Fresenius Environ. Bull. 2018, 27, 141–146. [Google Scholar]
- Yan, F.; Li, Z.J.; Dong, L.J.; Huang, R.; Cao, R.H.; Ge, J.; Xu, K.L. Cloud model-clustering analysis based evaluation for ventilation system of underground metal mine in alpine region. J. Cent. South Univ. 2021, 28, 796–815. [Google Scholar] [CrossRef]
- Lin, C.J.; Zhang, M.; Li, L.P.; Zhou, Z.Q.; Liu, S.; Liu, C.; Li, T. Risk Assessment of Tunnel Construction Based on Improved Cloud Model. J. Performence Constr. Facil. 2020, 34, 04020028. [Google Scholar] [CrossRef]
- Cong, X.H.; Ma, L. Performance Evaluation of Public-Private Partnership Projects from the Perspective of Efficiency, Economic, Effectiveness, and Equity: A Study of Residential Renovation Projects in China. Sustainability 2018, 10, 1951. [Google Scholar] [CrossRef] [Green Version]
- Li, G.; Zhang, Y.; Wang, H.C. Civil aviation master data recognition method based on cloud model and rough set. Comput. Eng. Des. 2020, 41, 2338–2344. [Google Scholar]
- Sun, Y.B.; Zhang, Y. Research on green development evaluation of coal listed companies based on cloud model. China Min. 2020, 29, 79–85. [Google Scholar]
- Liu, F.; Zhu, X.; Hu, Y.; Ren, L.; Johnson, H. A Cloud Theory-Based Trust Computing Model in Social Networks. Entropy 2017, 19, 11. [Google Scholar] [CrossRef] [Green Version]
- Lai, K. Service capability and performance of logistics service providers. Transp. Res. Part E Logist. Transp. Rev. 2004, 40, 385–399. [Google Scholar] [CrossRef]
- Ma, S.; Meng, Q. Research status and development trend of supply chain logistics capability. Comput. Integr. Manuf. Syst. 2005, 11, 301–307. [Google Scholar]
- Ming, W.; Hao, F. Research on the Development Policy of China’s Logistics Industry; China Planning Press: Beijing, China, 2002; p. 17. [Google Scholar]
- Jiang, J.; Liu, Z. Identification of three important capabilities of logistics enterprises. Logist. Technol. 2005, 7, 18–21. [Google Scholar]
Target Layer | First Level Indicator Layer | Secondary Indicator Layer | Three-Level Indicator Layer |
---|---|---|---|
Evaluation of regional logistics decarbonization development X | Low carbon logistics environment support force X1 | Economic environment | Gross regional product per capita X1,1 |
Fiscal revenue per capita X1,2 | |||
Total retail sales of social goods per capitaX1,3 | |||
Policy environment | The part of local financial expenses on environmental protection to total expenses X1,4 | ||
Logistics industry as a proportion of fixed investment X1,5 | |||
Low carbon logistics strength X2 | Logistics infrastructure | Road Density X2,1 | |
Rail Density X2,2 | |||
Logistics industry scale | Logistics operations per head X2,3 | ||
E-commerce sales per capita X2,4 | |||
Increase in logistics per capitaX2,5 | |||
The proportion of logistics employees in the workforce X2,6 | |||
Cargo turnover per capita X2,7 | |||
Logistics industry efficiency | Contribution of logistics industry to GDP X2,8 | ||
Value added of logistics industry per logistics employee X2,9 | |||
Carbon emissions per unit of added value in the logistics industry X2,10 | |||
Low carbon logistics potential X3 | Logistics industry input | Growth rate of new fixed asset investment in logistics industry X3,1 | |
Logistics workforce growth rate X3,2 | |||
Logistics output | Value added growth rate of logistics industry X3,3 | ||
Logistics industry demand | GDP per capita growth rateX3,4 | ||
Technical support | Technology Market Turnover Growth RateX3,5 | ||
R&D expenditure growth rate X3,6 |
Energy Name | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 |
---|---|---|---|---|---|---|---|
Raw Coal (million tons) | 15.86 | 16.03 | 12.30 | 7.97 | 3.22 | 0.94 | 0.41 |
Gasoline (million tons) | 45.40 | 46.45 | 44.65 | 41.62 | 42.41 | 42.57 | 49.74 |
Kerosene (million tons) | 476.51 | 507.07 | 543.78 | 593.66 | 643.31 | 690.47 | 697.17 |
Diesel (million tons) | 124.28 | 126.56 | 118.00 | 109.92 | 106.98 | 110.11 | 99.81 |
Fuel Oil (million tons) | 1.59 | 1.88 | 1.79 | 1.49 | 1.50 | 0.08 | 0.27 |
Liquefied Petroleum Gas (million tons) | 0.35 | 0.32 | 0.38 | 0.28 | 1.17 | 1.55 | 17.41 |
Natural Gas (billion kilowatt hours) | 2.35 | 3.17 | 2.11 | 1.99 | 1.80 | 3.72 | 3.42 |
Power (billion kilowatt hours) | 44.64 | 45.02 | 47.31 | 50.61 | 53.29 | 582.03 | 57.98 |
Energy Name | Discount Factor for Standard Coal | Unit | Carbon Emission Factor | Unit |
---|---|---|---|---|
Raw Coal | 0.7143 | million tons of standard coal/million tons | 0.7559 | Tonnes of carbon/tonne of standard coal |
Gasoline | 1.4714 | million tons of standard coal/million tons | 0.5538 | Tonnes of carbon/tonne of standard coal |
Kerosene | 1.4714 | million tons of standard coal/million tons | 0.5714 | Tonnes of carbon/tonne of standard coal |
Diesel | 1.4571 | million tons of standard coal/million tons | 0.5821 | Tonnes of carbon/tonne of standard coal |
Fuel Oil | 1.4286 | million tons of standard coal/million tons | 0.6185 | Tonnes of carbon/tonne of standard coal |
Liquefied Petroleum Gas | 1.7143 | million tons of standard coal/million tons | 0.5042 | Tonnes of carbon/tonne of standard coal |
Natural Gas | 13.3 | million tons of standard coal/billion cubic meters | 0.4483 | Tonnes of carbon/tonne of standard coal |
Power | 1.229 | million tons of standard coal/billion kilowatt hours | 2.2132 | Tonnes of carbon/tonne of standard coal |
Region | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 |
---|---|---|---|---|---|---|---|
Beijing | 688.7401 | 723.4680 | 743.4722 | 781.6607 | 825.9406 | 2315.8326 | 904.9456 |
Tianjin | 283.6145 | 302.9090 | 326.6324 | 339.6521 | 350.2898 | 362.5607 | 377.9557 |
Hebei | 724.9221 | 699.0801 | 519.9094 | 791.0719 | 762.9008 | 809.7153 | 1003.4259 |
Indicators | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 |
---|---|---|---|---|---|---|---|
X1,1 | 9.9927 | 10.6533 | 11.4137 | 12.4442 | 13.7646 | 15.3695 | 16.4555 |
X1,2 | 1.7310 | 1.8714 | 2.1759 | 2.3384 | 2.5015 | 2.6861 | 2.7006 |
X1,3 | 4.1948 | 4.4786 | 4.7619 | 5.0645 | 5.3318 | 6.6956 | 6.9934 |
X1,4 | 0.0331 | 0.0472 | 0.0529 | 0.0567 | 0.0672 | 0.0535 | 0.0417 |
X1,5 | 0.0966 | 0.1117 | 0.0960 | 0.0965 | 0.1359 | 0.1601 | 0.1389 |
X2,1 | 1.3207 | 1.3314 | 1.3336 | 1.3422 | 1.3544 | 1.3562 | 1.3629 |
X2,2 | 0.0778 | 0.0783 | 0.0783 | 0.0709 | 0.0770 | 0.0770 | 0.0833 |
X2,3 | 0.4023 | 0.4821 | 0.6281 | 0.5686 | 0.7341 | 1.1534 | 1.6162 |
X2,4 | 3.6093 | 4.2612 | 4.8934 | 5.5397 | 8.4610 | 8.4114 | 10.7873 |
X2,5 | 0.3171 | 0.3368 | 0.3408 | 0.3639 | 0.4150 | 0.4716 | 0.4693 |
X2,6 | 0.0798 | 0.0796 | 0.0772 | 0.0735 | 0.0710 | 0.0735 | 0.0746 |
X2,7 | 0.4970 | 0.4817 | 0.4152 | 0.3799 | 0.4415 | 0.4801 | 0.5058 |
X2,8 | 0.0317 | 0.0316 | 0.0299 | 0.0292 | 0.0302 | 0.0307 | 0.0285 |
X2,9 | 11.3277 | 12.0399 | 12.3300 | 13.5876 | 15.6153 | 16.8754 | 17.1322 |
X2,10 | −1.0271 | −0.9982 | −1.0050 | −0.9884 | −0.9167 | −2.2796 | −0.8953 |
X3,1 | −0.0568 | 0.1692 | −0.0691 | 0.0653 | 0.4825 | 0.1130 | −0.1540 |
X3,2 | 0.0242 | 0.0169 | −0.0033 | −0.0300 | −0.0086 | 0.0433 | −0.0199 |
X3,3 | 0.0552 | 0.0808 | 0.0207 | 0.0689 | 0.1394 | 0.1275 | −0.0050 |
X3,4 | 0.0854 | 0.0638 | 0.0669 | 0.0859 | 0.1050 | 0.1126 | 0.0749 |
X3,5 | 0.1599 | 0.1001 | 0.1010 | 0.1410 | 0.1385 | 0.1050 | 0.1487 |
X3,6 | 0.0796 | 0.0959 | 0.0453 | 0.0441 | 0.0559 | 0.0183 | 0.0408 |
Grade | Low | Lower | General | Higher | High |
---|---|---|---|---|---|
X1,1 | (0, 3) | (3, 6) | (6, 9) | (9, 13) | (13, 17) |
X1,2 | (0, 0.6) | (0.6, 1.2) | (1.2, 1.8) | (1.8, 2.4) | (2.4, 3) |
X1,3 | (0, 1) | (1, 2) | (2, 4) | (4, 6) | (6, 8) |
X1,4 | (0, 0.02) | (0.02, 0.03) | (0.03, 0.05) | (0.05, 0.06) | (0.06, 0.07) |
X1,5 | (0, 0.02) | (0.02, 0.05) | (0.05, 0.1) | (0.1, 0.15) | (0.15, 0.2) |
X2,1 | (0, 0.4) | (0.4, 0.8) | (0.8, 1.2) | (1.2, 1.6) | (1.6, 2) |
X2,2 | (0, 0.02) | (0.02, 0.04) | (0.04, 0.06) | (0.06, 0.08) | (0.08, 0.1) |
X2,3 | (0, 0.3) | (0.3, 0.6) | (0.6, 0.9) | (0.9, 1.3) | (1.3, 1.7) |
X2,4 | (0, 1) | (1, 3) | (3, 5) | (5, 8) | (8, 11) |
X2,5 | (0, 0.12) | (0.12, 0.24) | (0.24, 0.36) | (0.36, 0.48) | (0.48, 0.6) |
X2,6 | (0, 0.02) | (0.02, 0.04) | (0.04, 0.06) | (0.06, 0.08) | (0.08, 0.1) |
X2,7 | (0, 1) | (1, 3) | (3, 5) | (5, 7) | (7, 13) |
X2,8 | (0, 0.02) | (0.02, 0.04) | (0.04, 0.06) | (0.06, 0.08) | (0.08, 0.1) |
X2,9 | (0, 25) | (25, 50) | (50, 75) | (75, 100) | (100, 110) |
X2,10 | (−2.5, −1.5) | (−1.5, −1) | (−1, −0.6) | (−0.6, −0.3) | (−0.3, 0) |
X3,1 | (−1, 0) | (0, 0.1) | (0.1, 0.3) | (0.3, 0.4) | (0.5, 0.6) |
X3,2 | (−1, 0) | (0, 0.05) | (0.05, 0.1) | (0.1, 0.15) | (0.15, 0.2) |
X3,3 | (−1, 0) | (0, 0.05) | (0.05, 0.1) | (0.1, 0.15) | (0.15, 0.2) |
X3,4 | (−1, 0) | (0, 0.05) | (0.05, 0.1) | (0.1, 0.15) | (0.15, 0.2) |
X3,5 | (−1, 0) | (0, 0.5) | (0.5, 1) | (1, 2) | (2, 3) |
X3,6 | (−1, 0) | (0, 0.05) | (0.05, 0.1) | (0.1, 0.15) | (0.15, 0.2) |
Grade | Low | Lower | General | Higher | High |
---|---|---|---|---|---|
X1,1 | (1.5, 1.2739, 0.2) | (4.5, 1.2739, 0.2) | (7.5, 1.2739, 0.2) | (11, 1.6985, 0.3) | (15, 1.6985, 0.3) |
X1,2 | (0.3, 0.2548, 0.05) | (0.9, 0.2548, 0.05) | (1.5, 0.2548, 0.05) | (2.1, 0.2548, 0.05) | (2.7, 0.2548, 0.05) |
X1,3 | (0.5, 0.4246, 0.1) | (1.5, 0.4246, 0.1) | (3, 0.8493, 0.15) | (5, 0.8493, 0.15) | (7, 0.8493, 0.15) |
X1,4 | (0.01, 0.0085, 0.0015) | (0.025, 0.0042, 0.001) | (0.04, 0.0085, 0.0015) | (0.055, 0.0042, 0.001) | (0.065, 0.0042, 0.001) |
X1,5 | (0.01, 0.0085, 0.0015) | (0.035, 0.0127, 0.002) | (0.075, 0.0212, 0.003) | (0.125, 0.0212, 0.003) | (0.175, 0.0212, 0.003) |
X2,1 | (0.2, 0.1699, 0.03) | (0.6, 0.1699, 0.03) | (1, 0.1699, 0.03) | (1.4, 0.1699, 0.03) | (1.8, 0.1699, 0.03) |
X2,2 | (0.01, 0.0085, 0.0015) | (0.03, 0.0085, 0.0015) | (0.05, 0.0085, 0.0015) | (0.07, 0.0085, 0.0015) | (0.09, 0.0085, 0.0015) |
X2,3 | (0.15, 0.1274, 0.02) | (0.45, 0.1274, 0.02) | (0.75, 0.1274, 0.02) | (1.1, 0.1699, 0.03) | (1.5, 0.1699, 0.03) |
X2,4 | (0.5, 0.4246, 0.1) | (2, 0.8493, 0.15) | (4, 0.8493, 0.15) | (6.5, 1.2739, 0.2) | (9.5, 1.2739, 0.2) |
X2,5 | (0.06, 0.0510, 0.01) | (0.18, 0.0510, 0.01) | (0.3, 0.0510, 0.01) | (0.42, 0.0510, 0.01) | (0.54, 0.0510, 0.01) |
X2,6 | (0.01, 0.0085, 0.0015) | (0.03, 0.0085, 0.0015) | (0.05, 0.0085, 0.0015) | (0.07, 0.0085, 0.0015) | (0.09, 0.0085, 0.0015) |
X2,7 | (0.5, 0.4246, 0.1) | (2, 0.8493, 0.15) | (4, 0.8493, 0.15) | (6, 0.8493, 0.15) | (10, 2.5478, 0.4) |
X2,8 | (0.01, 0.0085, 0.0015) | (0.03, 0085, 0.0015) | (0.05, 0.0085, 0.0015) | (0.07, 0.0085, 0.0015) | (0.09, 0.0085, 0.0015) |
X2,9 | (12.5, 10.6157, 1.8) | (37.5, 10.6157, 1.8) | (62.5, 10.6157, 1.8) | (87.5, 10.6157, 1.8) | (105, 4.2463, 0.7) |
X2,10 | (−2, 0.4246, 0.1) | (−1.25, 0.2123, 0.03) | (−0.8, 0.1699, 0.03) | (−0.45, 0.1274, 0.02) | (−0.15, 0.1274, 0.02) |
X3,1 | (−0.5, 0.4246, 0.1) | (0.05, 0.0425, 0.01) | (0.2, 0.0850, 0.015) | (0.35, 0.0425, 0.01) | (0.55, 0.0425, 0.01) |
X3,2 | (−0.5, 0.4246, 0.1) | (0.025, 0.0212, 0.003) | (0.075, 0.0212, 0.003) | (0.125, 0.0212, 0.003) | (0.175, 0.0212, 0.003) |
X3,3 | (−0.5, 0.4246, 0.1) | (0.025, 0.0212, 0.003) | (0.075, 0.0212, 0.003) | (0.125, 0.0212, 0.003) | (0.175, 0.0212, 0.003) |
X3,4 | (−0.5, 0.4246, 0.1) | (0.025, 0.0212, 0.003) | (0.075, 0.0212, 0.003) | (0.125, 0.0212, 0.003) | (0.175, 0.0212, 0.003) |
X3,5 | (−0.5, 0.4246, 0.1) | (0.25, 0.2123, 0.03) | (0.75, 0.2123, 0.03) | (1.5, 0.4246, 0.1) | (2.5, 0.4246, 0.1) |
X3,6 | (−0.5, 0.4246, 0.1) | (0.025, 0.0212, 0.003) | (0.075, 0.0212, 0.003) | (0.125, 0.0212, 0.003) | (0.175, 0.0212, 0.003) |
Grade | Low | Lower | General | Higher | High | Grade |
---|---|---|---|---|---|---|
X1,1 | 0.0000 | 0.0006 | 0.1504 | 0.8250 | 0.0222 | Higher |
X1,2 | 0.0000 | 0.0122 | 0.6359 | 0.3419 | 0.0034 | General |
X1,3 | 0.0000 | 0.0000 | 0.3626 | 0.6251 | 0.0095 | Higher |
X1,4 | 0.0360 | 0.1714 | 0.7032 | 0.0002 | 0.0000 | General |
X1,5 | 0.0000 | 0.0001 | 0.5851 | 0.4010 | 0.0027 | General |
X2,1 | 0.0000 | 0.0009 | 0.1732 | 0.8858 | 0.0276 | Higher |
X2,2 | 0.0000 | 0.0000 | 0.0104 | 0.6376 | 0.3473 | Higher |
X2,3 | 0.1492 | 0.9276 | 0.0324 | 0.0013 | 0.0000 | Lower |
X2,4 | 0.0000 | 0.1711 | 0.8907 | 0.0850 | 0.0002 | General |
X2,5 | 0.0001 | 0.0411 | 0.9385 | 0.1421 | 0.0009 | General |
X2,6 | 0.0000 | 0.0000 | 0.0060 | 0.4999 | 0.4717 | Higher |
X2,7 | 1.0000 | 0.2158 | 0.0013 | 0.0000 | 0.0028 | low |
X2,8 | 0.0501 | 0.9780 | 0.1101 | 0.0005 | 0.0000 | Higher |
X2,9 | 0.9933 | 0.0598 | 0.0002 | 0.0000 | 0.0000 | low |
X2,10 | 0.0888 | 0.5643 | 0.3939 | 0.0003 | 0.0000 | Higher |
X3,1 | 0.5553 | 0.0633 | 0.0191 | 0.0000 | 0.0000 | low |
X3,2 | 0.4427 | 0.9992 | 0.0651 | 0.0001 | 0.0000 | Higher |
X3,3 | 0.4102 | 0.3580 | 0.6335 | 0.0079 | 0.0000 | General |
X3,4 | 0.3705 | 0.0239 | 0.8801 | 0.1793 | 0.0006 | General |
X3,5 | 0.2948 | 0.9096 | 0.0285 | 0.0179 | 0.0001 | Higher |
X3,6 | 0.3812 | 0.0439 | 0.9752 | 0.1069 | 0.0003 | General |
C | 0.2233 | 0.2623 | 0.3704 | 0.2311 | 0.0322 | General |
Indicators | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | Weights |
---|---|---|---|---|---|---|---|---|
X1,1 | 0.0000 | 0.1022 | 0.2199 | 0.3793 | 0.5836 | 0.8320 | 1.0000 | 0.0513 |
X1,2 | 0.0000 | 0.1448 | 0.4588 | 0.6264 | 0.7947 | 0.9850 | 1.0000 | 0.0394 |
X1,3 | 0.0000 | 0.1014 | 0.2026 | 0.3108 | 0.4063 | 0.8936 | 1.0000 | 0.0572 |
X1,4 | 0.0000 | 0.4135 | 0.5806 | 0.6921 | 1.0000 | 0.5982 | 0.2522 | 0.0333 |
X1,5 | 0.0094 | 0.2449 | 0.0000 | 0.0078 | 0.6225 | 1.0000 | 0.6693 | 0.0886 |
X2,1 | 0.0000 | 0.2536 | 0.3057 | 0.5095 | 0.7986 | 0.8412 | 1.0000 | 0.0376 |
X2,2 | 0.5565 | 0.5968 | 0.5968 | 0.0000 | 0.4919 | 0.4919 | 1.0000 | 0.0270 |
X2,3 | 0.0000 | 0.0657 | 0.1860 | 0.1370 | 0.2733 | 0.6187 | 1.0000 | 0.0714 |
X2,4 | 0.0000 | 0.0908 | 0.1789 | 0.2689 | 0.6759 | 0.6690 | 1.0000 | 0.0566 |
X2,5 | 0.0000 | 0.1275 | 0.1534 | 0.3029 | 0.6337 | 1.0000 | 0.9851 | 0.0574 |
X2,6 | 1.0000 | 0.9773 | 0.7045 | 0.2841 | 0.0000 | 0.2841 | 0.4091 | 0.0399 |
X2,7 | 0.9301 | 0.8086 | 0.2804 | 0.0000 | 0.4893 | 0.7959 | 1.0000 | 0.0320 |
X2,8 | 1.0000 | 0.9688 | 0.4375 | 0.2188 | 0.5313 | 0.6875 | 0.0000 | 0.0366 |
X2,9 | 0.0000 | 0.1227 | 0.1727 | 0.3893 | 0.7387 | 0.9558 | 1.0000 | 0.0535 |
X2,10 | 0.9048 | 0.9257 | 0.9208 | 0.9327 | 0.9845 | 0.0000 | 1.0000 | 0.0221 |
X3,1 | 0.1527 | 0.5078 | 0.1334 | 0.3445 | 1.0000 | 0.4195 | 0.0000 | 0.0526 |
X3,2 | 0.7394 | 0.6398 | 0.3643 | 0.0000 | 0.2920 | 1.0000 | 0.1378 | 0.0450 |
X3,3 | 0.4169 | 0.5942 | 0.1780 | 0.5118 | 1.0000 | 0.9176 | 0.0000 | 0.0388 |
X3,4 | 0.4426 | 0.0000 | 0.0635 | 0.4529 | 0.8443 | 1.0000 | 0.2275 | 0.0547 |
X3,5 | 1.0000 | 0.0000 | 0.0151 | 0.6839 | 0.6421 | 0.0819 | 0.8127 | 0.0668 |
X3,6 | 0.7899 | 1.0000 | 0.3479 | 0.3325 | 0.4845 | 0.0000 | 0.2899 | 0.0380 |
Grade | Low | Lower | General | Higher | High | Evaluation Results |
---|---|---|---|---|---|---|
2013 | 0.2204 | 0.2689 | 0.3709 | 0.2260 | 0.0315 | Genera |
2014 | 0.2011 | 0.2365 | 0.3887 | 0.2838 | 0.0347 | Genera |
2015 | 0.2199 | 0.2262 | 0.2756 | 0.3306 | 0.0398 | Higher |
2016 | 0.2052 | 0.2422 | 0.2594 | 0.3657 | 0.0530 | Higher |
2017 | 0.1812 | 0.1463 | 0.2054 | 0.3581 | 0.1863 | Higher |
2018 | 0.2093 | 0.1874 | 0.1329 | 0.3762 | 0.2357 | Higher |
2019 | 0.2202 | 0.1678 | 0.1777 | 0.1642 | 0.2708 | High |
Grade | Low | Lower | General | Higher | High | Evaluation Results |
---|---|---|---|---|---|---|
2013 | 0.2440 | 0.3131 | 0.2515 | 0.2546 | 0.0808 | Lower |
2014 | 0.2166 | 0.3202 | 0.2887 | 0.2628 | 0.0486 | Lower |
2015 | 0.1852 | 0.3268 | 0.2735 | 0.2599 | 0.0594 | Lower |
2016 | 0.2378 | 0.3301 | 0.2310 | 0.2435 | 0.0656 | Lower |
2017 | 0.1437 | 0.3075 | 0.3556 | 0.2153 | 0.0661 | General |
2018 | 0.1284 | 0.3849 | 0.3815 | 0.1642 | 0.0623 | Lower |
2019 | 0.0741 | 0.2191 | 0.3531 | 0.2450 | 0.1517 | General |
Grade | Low | Lower | General | Higher | High | Evaluation Results |
---|---|---|---|---|---|---|
2013 | 0.2859 | 0.2655 | 0.2323 | 0.1925 | 0.1081 | Low |
2014 | 0.3137 | 0.3102 | 0.2573 | 0.1368 | 0.0664 | Low |
2015 | 0.2849 | 0.3590 | 0.2781 | 0.1271 | 0.0834 | Lower |
2016 | 0.2826 | 0.2778 | 0.3544 | 0.1294 | 0.0526 | General |
2017 | 0.2692 | 0.2677 | 0.3071 | 0.1574 | 0.1307 | General |
2018 | 0.1639 | 0.3536 | 0.3213 | 0.1601 | 0.1493 | Lower |
2019 | 0.1619 | 0.2760 | 0.3480 | 0.2277 | 0.1552 | General |
Indicators | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 |
---|---|---|---|---|---|---|---|
X1,1 | Higher | Higher | Higher | Higher | High | High | High |
X1,2 | General | Higher | Higher | Higher | High | High | High |
X1,3 | Higher | Higher | Higher | Higher | Higher | High | High |
X1,4 | General | General | Higher | Higher | High | Higher | General |
X1,5 | General | General | General | General | General | General | General |
X2,1 | Higher | Higher | Higher | Higher | Higher | Higher | Higher |
X2,2 | Higher | Higher | Higher | Higher | Higher | Higher | High |
X2,3 | Lower | Lower | General | Lower | General | Higher | High |
X2,4 | General | General | General | Higher | High | High | High |
X2,5 | General | General | General | Higher | Higher | Higher | Higher |
X2,6 | Higher | Higher | Higher | Higher | Higher | Higher | Higher |
X2,7 | Low | Low | Low | Low | Low | Low | Low |
X2,8 | Lower | Lower | Lower | Lower | Lower | Lower | Lower |
X2,9 | Low | Low | Low | Low | Low | Low | Low |
X2,10 | Lower | General | Lower | General | General | Low | General |
X3,1 | Low | General | Low | Lower | High | General | Low |
X3,2 | Lower | Lower | Low | Low | Low | Lower | Low |
X3,3 | General | General | Lower | General | Higher | Higher | Low |
X3,4 | General | General | General | General | Higher | Higher | General |
X3,5 | Lower | Lower | Lower | Lower | Lower | Lower | Lower |
X3,6 | General | General | Lower | Lower | General | Lower | Lower |
C | General | General | Higher | Higher | Higher | Higher | High |
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Guo, Z.; Tian, Y.; Guo, X.; He, Z. Research on Measurement and Application of China’s Regional Logistics Development Level under Low Carbon Environment. Processes 2021, 9, 2273. https://doi.org/10.3390/pr9122273
Guo Z, Tian Y, Guo X, He Z. Research on Measurement and Application of China’s Regional Logistics Development Level under Low Carbon Environment. Processes. 2021; 9(12):2273. https://doi.org/10.3390/pr9122273
Chicago/Turabian StyleGuo, Zixue, Yu Tian, Xinmei Guo, and Zefang He. 2021. "Research on Measurement and Application of China’s Regional Logistics Development Level under Low Carbon Environment" Processes 9, no. 12: 2273. https://doi.org/10.3390/pr9122273
APA StyleGuo, Z., Tian, Y., Guo, X., & He, Z. (2021). Research on Measurement and Application of China’s Regional Logistics Development Level under Low Carbon Environment. Processes, 9(12), 2273. https://doi.org/10.3390/pr9122273